Most teams pick a data provider and hope it works. You can actually test every option and see what works for your business. That is what we do at Deepline, and it is simpler than people think.
Video not loading in LinkedIn or your in-app browser? Watch it on YouTube.
One API, every provider
Sales teams always ask, "what's the best data provider?" My answer never changes. It depends. You have to test it for yourself. Match rates vary by provider mix, vertical, company size, and geography, so the right answer is always specific to you.
Before, this took forever.
"As recently as October 2025 he had to manually go through this process of building a propensity model for one of our clients. It took four weeks. We essentially manually enriched all these accounts, 15,000 accounts... today we were able to do it in about 30 minutes."
We built Deepline so you don't have to waste weeks guessing. You get one API, one SDK, one billing profile. You can try hundreds of providers with the same credits. Your agents can now programmatically test a hundred different variations to actually solve your problem.
The explore-exploit pattern
What works for you won't work for everyone else. You have to explore all the options, then double down on the ones that actually work. That is the classic explore-exploit tradeoff from reinforcement learning. We call it programmatic GTM. You give the system a sample of good outcomes (like closed-won accounts), and let it find the patterns.
"The model that just keeps working is you define the outcome of what good looks like. You let the systems actually programmatically find the patterns."
Demo: finding the best waterfall
Here's how it works. We ran Deepline's pre-research tool on a list of restaurant accounts. It checked social channels, scraped websites, tried paid providers, pulled public and private data, and generated a report. Four sources accounted for 60% of the wins. They were cheap and fast. We did it in under 30 minutes.
Turn every process into code
Every go-to-market process can be scripted. You get a TypeScript script for every workflow. It's easier to review, backtest, and update than clicking around in a tool. This is what GTM engineering is really about: applying software-engineering discipline to go-to-market work instead of running one-off manual processes. You spend less time spot checking, more time shipping.
The trick is to show the system what good looks like, then let it try every path to get there. Combine public and private data. Backtest. You'll find results that surprise you. Every customer finds something new and specific to their business.
Build your own tools
Don't settle for generic answers. Build your own tools. Programmatic GTM means less manual work and results that actually matter for your team. If you want to try it, grab the Deepline CLI. It takes two minutes to set up. If you're an engineer who likes building weird stuff, we're hiring.
Part of the GTM + AI NYC Lightning Talks - see all six talks. Hosted by Deepline at Ramp HQ.
Jai Toor on LinkedIn · Deepline
More from the event: Keyan (Ramp) · Bryant (OpenAI) · Julia (Notion) · Chris (Modal) · Jacob (Attention)